23:24 <nicksydney> DocScrutinizer05: http://faif.us/cast/2014/jul/18/0x49/

23:24 <nicksydney> [Get cast in Ogg/Vorbis format] [Get cast in MP3 format] 0x49: Why Free Software Phone Doesn't Exist

23:29 <DocScrutinizer05> so?

23:30 <DocScrutinizer05> I doubt I will learn anything new (which I also would be willing to believe) from such a talk

23:31 <DocScrutinizer05> I've been *inside* business of development modem firmware, I think I know why there never will be a Free Software Phone

23:32 <DocScrutinizer05> nobody can pay for, nor assure integrity of the mandatory certifications

23:33 <wpwrak> time for the roguephone ;-)

23:33 <whitequark> time to kill cellular as a technology

23:34 <DocScrutinizer05> when a company like ST-E employs 1200 engineers for 24 months to develop a LTE firmware, why would they want to put it into public domain?

23:34 <whitequark> every time you tell that story, the amount of engineers and months grows :D

23:35 <wpwrak> maye we should talk to the fine folks at ISIS, or make friends in north korea. they may be less reluctant to ignore western restrictions :)

23:35 <DocScrutinizer05> depends what you count in

23:36 <DocScrutinizer05> wpwrak: sure, and they have an abundance of highly qualified software engineers

23:37 <wpwrak> consider it a job opportunity ;-)

23:39 <wpwrak> man ! that kinetis oscillator sucks !

23:40 <DocScrutinizer05> where will they build the 60nm chip fab?

23:40 <wpwrak> they will _occupy_ it ;-)

23:41 <DocScrutinizer05> yeah, and then cut into little pieces and ship from Dresden to North Iraq, eh?

23:42 <wpwrak> how to crash a KL26, method #651: OSC0_CR |= whatever; // change some harmless-looking field in that oscillator control register. this will read a "1" from some "unused" fields. that same "unused" fields really hates it when a "1" is written to it ...

23:42 <DocScrutinizer05> ooooold

23:43 <whitequark> uhhh, never touch "unused" fields

23:43 <wpwrak> or at least i suppose there's a "1" somewhere in there. 'cause the whole critter goes south if i use |=. it's happy enough to accept everything i send it if i use =

23:43 <whitequark> microcontrollers 101

23:43 <DocScrutinizer05> man i2ctools

23:43 <wpwrak> whitequark: tjey'

23:43 <wpwrak> lert's try this again

23:44 <wpwrak> they're supposed to read back a "0" ...

23:44 <DocScrutinizer05> never do bit operation directly on control registers

23:45 <whitequark> you can do that on some architectures

23:45 <whitequark> e.g. STM32 specifically supports it, and most cortex SoCs do

23:45 <DocScrutinizer05> depends on your ALU and how it works

23:45 <whitequark> it's more due to the way register responds to writes

23:46 <whitequark> especially flags that get reset on writes

23:46 <DocScrutinizer05> well, kinda, but also how CPU does a read-modify-write

23:47 <whitequark> yeah, there's some wankery with byte- and word-sized reads and writes

23:47 <whitequark> and peripherals that react to reads/writes directly

23:47 <whitequark> and the way your bus divides/groups requests

23:47 <DocScrutinizer05> yep

23:47 <wpwrak> i do it with perfect impunity on any other register so far ...

23:47 <whitequark> anyway, cortex-m socs are explicitly designed to be completely compatible with C

23:47 <whitequark> you can even do interrupt handlers without a line of assembly

23:48 <wpwrak> yup, and this is an M0+

23:48 <whitequark> because its interrupt controller knows C ABI

23:48 <whitequark> (it's also faster because of that, as it can switch to an interrupt of higher priority without repushing registers)

23:48 <whitequark> (if it arrives while pushing them)

23:48 <wpwrak> well, in a bit, i'll make my code return that register value. then i'll know for sure what's cooking there.

23:48 <wpwrak> for now, all i gave is a lone LED :)

23:49 <wpwrak> s/gave/have/

23:49 <whitequark> hook a scope there

23:49 <whitequark> then shift it out

23:52 <wpwrak> also found that my rfkill switch kills a bit more than just RF - i flipped VRF and VSYS and didn't notice. so it either connects VRF to VSYS (rf enabled, all is well), or VSYS to GND (system not happy)

23:53 <wpwrak> ah well, that's why we have sharp knives and those yellow wires ...

23:54 <wpwrak> mean while, i'm learning the dark secrets of the FLL. seems that it runs a tad too fast ...

23:54 <whitequark> fll?

23:54 <whitequark> frequency-locked loop?

23:55 <wpwrak> yup

23:55 <wpwrak> the chip has PLL and FLL .. and some tricky rules when to use which

23:58 <wpwrak> the whole clocking system is actually a mess. it consists of no less than five subsystems with documentation on a module basis. so the word "external" can have a great many meanings.

23:58 <wpwrak> and you have a while army of clock gates and clock mode bits

23:58 <wpwrak> and of course no overview diagram that would give you the whole picture. just one that at least tells you where some of the muxers are lurking.

23:59 <wpwrak> just enough information that you can eventually figure things out within a small number of days of trial and error ...